The provision of healthcare has long been a topic of global concern, yet when medical infrastructure is pushed to its limits – as it is with the current pandemic - our ability to respond has chiefly relied upon speed of human reaction. However, technological developments, particularly in the field of Artificial Intelligence, have the potential to drastically redress the imbalance between supply and demand of healthcare, and improve the analysis and quality of treatment that patients receive.
When we think of Artificial Intelligence, we tend to think of highly intelligent robots or computers that would pass the Turing Test – that is, you cannot distinguish them from a human. This is sometimes called ‘Strong AI’ and is probably decades from becoming a reality. The good news is that lesser forms of AI such as machine learning and deep learning are here now – and they are already having an impact on healthcare, even being used to help manage the coronavirus outbreak.
Examples of current applications include facial recognition systems that can detect the symptoms of fever, sophisticated monitoring equipment, AI chatbots for online consultations and software for pandemic modelling. But what about healthcare more generally? What possibilities does AI hold out for a healthier world once we get through the current crisis?
Greater productivity and reduced costs
Martin Ciupa is an expert in AI. He is CEO of the start-up Mental Imagery, CTO of Kimbocare, a blockchain-enabled tech platform, and has served as chief AI officer at Mindmaze, a Swiss unicorn that builds intuitive human machine interfaces. Recently, he was also elected leader of the American Association of Precision Medicine’s Data Science Coronavirus Taskforce. He believes there are four main areas where AI will improve medicine; productivity, quality, personalisation, and discovery.
The first, productivity, is about making healthcare more efficient. “AI can relieve pressures on healthcare and also augment people, allowing them to do more work in a given day,” says Ciupa. This is a potentially huge development. To take the first statement, according to the Harvard Business Review, labour is the single biggest cost in US healthcare. Moreover, productivity has been declining for decades and it is estimated that of the $3trn spent annually on healthcare one third is wasted.
This wastage is often most visible in areas like poor bed allocation and other administrative inefficiencies. But it also manifests itself in having expensive people, such as doctors doing administrative or mundane medical tasks, the kind of things AI excels at. Use it to replace people and not only do you save money but doctors can spend more time with patients.
When it comes to productivity it is worth remembering that healthcare is very different to other industries. In healthcare systems across the world, demand consistently outstrips supply (or affordable supply). Thus, if you can provide more service for the same expenditure you are likely to have a healthier population, not greater numbers of unemployed doctors or nurses. People will ‘consume’ more healthcare and the world will be a better, fitter place for it.
Increase efficiency to improve quality
The second area is quality. “An example of this is examining X-rays and MRI scans,” says Ciupa. “Deep learning is really good at looking at medical images and already performs at similar radiological levels as human experts.”
Crucially, this algorithm was reading them for 14 different pathologies, not just one, representing huge potential efficiency gains. Technology could even lead to digital consultations; Babylon, the med-tech company, has claimed that its chatbots can already surpass doctors in terms of the accuracy of diagnosis.
The benefits of automating routine tasks such as initial consultations range from vastly reduced waiting times in the developed world to access to medical services for the first time in the developing world. Huge strides are already been made with telemedicine in Africa. The use of AI to digest huge amounts of data could result in other discoveries too. “There’s a trial at London’s Moorfields Eye Hospital where AI is being used to look at thousands of eye scans,” says Kim Nilsson, CEO of Pivigo, a data science marketplace and training company. The scans are compared and cross-referenced with an individual’s medical history to see if they can show that patients are at risk of other conditions. “For instance, does your scan indicate you are at risk of diabetes?”
Nilsson adds that the use of AI to analyse and interpret huge amounts of data will become ever more important as the Internet of Things takes off and smart devices proliferate. With wearables that constantly stream information on your health, people could be monitored in real time and AI could pick up problems such as heart conditions while they are minor and easily treatable – and even before any conditions develop.
Truly personalised healthcare
While AI can discern patterns in the data of millions of patients, it can also make treatment on an individual basis much more personal. This will result in better outcomes and lower costs because people receive treatment tailored specifically to them. In a recent note, Giulia Besana, an analyst at IDC Health Insights wrote, “European healthcare providers are betting on AI to support greater personalisation of healthcare services.” But what does this mean in practice?
If you think of current healthcare as Healthcare 3.0, it’s set up a bit like one of Henry Ford’s factories.
“You come into hospital and get on a production line. You are given a standard battery of tests until the hospital discovers what is wrong with you. However,” he explains, “AI will result in Healthcare 4.0. This, like Industry 4.0 more generally, will mean greater automation, cyberphysical systems, data exchange and proliferation of AI.
For patients this will deliver a far more tailored, personalised service from the word go. The business of healthcare will become more efficient, and far more factors will be taken into account during consultation and treatment. “For example, a person who has Alzheimer’s may also have depression and anxiety,” says Ciupa. “At the moment these often cannot be taken into account when treating it.” Nor is it just the obvious stuff. AI might mean tailored, virtual reality physiotherapy for example. Again, this would be likely to deliver better, faster and cheaper results.
The last and perhaps most exciting area where AI will be a game-changer is the discovery of new drugs. “Developing a new drug can cost $1bn,” says Nilsson. “AI could cut that in half.” It can do this in all sorts of ways. The obvious ones perhaps are sequencing the DNA or RNA of new viruses and mapping them to suggest which treatments are likely to be most effective. Others include screening and monitoring participants in medical trials.
One of the most promising possibilities is repurposing existing drugs to treat new conditions. Existing drugs do not need the same sort of clinical trials; their side effects are known, they are already FDA or EMA approved, they are cheaper and more efficient. “You might look at an existing drug for arthritis which also has possibilities for treating high blood pressure,” says Nilsson.
More data, more insights
Ultimately, the future that AI and associated technologies holds for medicine is one of vastly more data and vastly more useful knowledge gleaned from that data. “You will go about your daily business while wearing your watch and your smartphone,” says Ciupa. “The data from these devices will allow AI to detect problems and recommend personalised treatments before you know anything is wrong.”
You will suffer from fewer illnesses and you will recover quicker. Indeed, some, such as Babylon’s Dr Ali Parsa, have predicted that AI and associated technologies will be so revolutionary for healthcare that in a decade or so the issue of funding will no longer be a problem.
Download the Investment Guide
Which topics will be relevant in the upcoming three months? Julius Baer experts share their veiws on the economy and financial markets.